Public Access to Research Data Used in Rule Making
During the workshop, a panel moderated by Alan Morrison and composed of individuals representing various interests (Bruce Alberts, National Academy of Sciences; Wendy Baldwin, National Institutes of Health; William H. Farland, U.S. Environmental Protection Agency; David G. Hawkins, Natural Resources Defense Fund; Jim J. Tozzi, Center for Regulatory Effectiveness; and William L. Kovacs, U.S. Chamber of Commerce) was asked to consider the following questions and to offer their opinions:
Is the primary purpose of the Shelby Amendment to strengthen the public’s “right to know” how tax dollars are spent, or is it to create a “right to see and contest data” that are used to support federal decision making?
Should access to data be limited to data used for formal rule making, or should there be access to data used for general policy making?
The Shelby Amendment does not set a timetable for making data available. Should the data be available when a rule is proposed, or when a final rule is issued?
Who should bear the cost of responding to a Shelby Amendment request? The agency? The grantee? The requester?
In what form should the data be delivered to the public? Who should bear responsibility for putting it into this form?
FOIA treats all requesters equally. Should there be any limits on who may request research data?
The careers of researchers often depend on the continuing use of already collected data. Must they give up such data if it means jeopardizing their careers?
Should the public have access to data funded and produced by for-profit entities who oppose a regulation, such as an industry coalition?
Responses to these questions are presented under the identified respondents.
The perspective of a bench scientist (Bruce Alberts-National Academy of Sciences). Science is a community effort; it succeeds only if other scientists can analyze, interpret, and extend each other’s work. Science requires full disclosure and availability to other scientists of the research methods, results, and, when necessary, special materials used in research. For example, Dr. Alberts noted the Human Genome Project requires extensive sharing not only of data, but also of materials, such as DNA clones and human cell lines. Virtually all scientific conclusions are based on a body of evidence, a complex web of studies related to one another in subject matter and reasoning.
While the academic community believes that the OMB did a thoughtful job in narrowing the Shelby Amendment, he said that concerns still exist. For example, the Shelby Amendment includes no “need to know” provision; anyone can request data for any reason. Without some front-end filtering mechanism, there is a danger that the amendment could be used to harass scientists whose work is found objectionable by anyone, for any reason. The potential for harassment is already real in some fields, and additional disincentives could further discourage the best young people from choosing careers in science and thus jeopardize the United States’ leadership position. Therefore, while sharing data with those outside the laboratory environment is a good and necessary idea, bench scientists believe that FOIA is not the best mechanism to achieve this.
The perspective of a funding agency (Wendy Baldwin-NIH). Over the past 10 to 20 years, the norms of data sharing have shifted dramatically, said Dr. Baldwin. Previously, investigators kept very large data sets virtually until they became historical records. Today, they have started to put data into the public domain even before they have published their findings based on the data.
There are several purposes to this custom of data sharing. As the NIH representative said, it “supports and forces scientific inquiry.” Data sharing allows researchers to make maximum use of the very large investments needed to create complex data sets. Additionally, data sets are effective vehicles for teaching the next generation of scientists. They help scientists develop new methodologies and a diversity of analytical views.
Data sharing also allows conclusions to be tested by people with different outlooks, disciplines, or statistical techniques.
FOIA, however, was not designed to force the sharing of data sets. She said that it was designed to make available certain government records, normally in written form, to members of the general public who can read and understand them. Scientific data are shared for the purpose of re-analysis—a scientific activity, not a public activity. Although it is easy to understand the appeal of gaining access to “original data,” this access can serve little purpose for those without the skills to reanalyze it.
Another problem raised by using FOIA is that the Act imposes no limits on how accessed information might be used. In a typical data-sharing environment, the researcher is allowed to put some restrictions on data use. One restriction might be against using data for non-research purposes, such as identifying a market segment or contacting people who took part in a study.
Under FOIA, someone besides the investigator strips away the “identifiable” personal information. However, it is the investigator who is in the best position to understand how the combinations of variables might be used to identify individuals. Deleting only a name, Social Security number and address may not be sufficient to protect the identities of research subjects who might object to having their personal information released. When awards are made to institutions, those institutions have the obligation to protect the confidentiality of the data gathered under the award.
To facilitate compliance with the Shelby Amendment, the NIH has begun to encourage investigators to put data into public archives, data enclaves, and other mechanisms that allow access for research purposes. Furthermore, it has begun a review and revision of informed consent forms and research on human subject clearances and procedures. The revised forms would alert individuals participating in human subjects research to the possible implications of the Shelby Amendment.
Failure to comply with the amendment may be deemed a violation of the terms and conditions of an NIH award. However, no mechanism is in place to handle disputes over access to the data. Such resolutions are potentially complex, because there are four potential parties: the principal investigator, the home institution, the funding agency, and the requestor. It is likely that differences in opinion will arise not only between a scientist and a requestor, but also between scientists and their home institutions.
The perspective of a regulatory agency (William H. Farland-EPA). Dr. Farland said that scientific data and studies used by the EPA move through a process designed to help decision makers. In doing risk analysis, the EPA’s job is to seek the best available data to reduce uncertainty
with regard to a particular risk. Most important, the agency tries to describe at the outset of a regulatory process how scientific information supports the decision, and to communicate the impact of that information. This is especially important when the debates among epidemiologists or statisticians confuse the public, who do not know which side to believe and who may lose faith in both the scientific process and the regulatory process.
The Shelby Amendment, Farland said, raises several questions for the EPA about rule making as a legal and deliberative process. At what point should the agency disclose what type of regulation is going to be considered or issued? The timing of the release can influence its reception. Should the agency use contracts to support the research needed for regulations? Contracting, as opposed to grants that support more flexible work, might narrow the type of information the agency receives and could possibly limit the scope of the science underlying the regulation.
In 1999, the EPA established a new Office for Environmental Information, which tries to provide environmental decision makers, both inside and outside government, with information needed to protect public health and the environment. The volume of information involved is large. The agency receives information from several sources: (1) in-house research programs, (2) contracts and grants, (3) routine or compliance monitoring data, (4) the regulatory process and the regulated industries, (5) rules from other agencies, (6) voluntarily submitted data, and (7) the open literature. For data generated under grants agreements, the EPA is already required to follow FOIA procedures. This also is true for data generated by agency researchers or contractors as part of in-house programs, if the contractor delivers the data to the agency.
The perspective of a public interest advocacy group (David G. Hawkins-Natural Resources Defense Council). Openness and transparency improve government actions and are necessary for democracy, said Mr. Hawkins. However, the Shelby Amendment may not yield these results.
First, application of the Shelby Amendment is “one-sided” because it applies only to federally funded research. “If access to the basic information upon which agencies make their decisions is the issue,” he said, “there is no principled basis for saying that access should be available only when the research has been federally funded.” For example, public advocacy groups do not have the right to use FOIA to demand the underlying data that may be present in industry-supported studies that have been submitted on a confidential basis to an agency to assert a claim of economic harm.
Second, he said, the Shelby Amendment increases the burden on an agency to justify its judgment that a hazard exists that warrants precautions. If affected interests can easily use the Shelby-modified OMB rules
to invalidate studies, federal agencies will have no basis for moving forward toward improved regulation. In other words, how does the government responsibly use scientific information that by its nature always contains uncertainties and ambiguities to describe potentially legitimate issues of hazards to public health and the environment?
One alternative to the Shelby Amendment, suggested by the NRDC representative, is to challenge the data used in rule making under the Administrative Procedure Act (APA). The APA, which is 50 years old, addresses legitimate issues about the reliability of data. For example, an agency that bases an action either on its own data or on a published study is responsible for responding to comments about the validity of the data. If the study design, or performance, or conclusion are said to raise any doubts, the agency has a duty under the APA to respond to such comments.
In addition, Hawkins said, the federal courts are experienced in evaluating the reasonableness of agency responses to public comment. For example, an agency might receive a request from a study participant or a study from another interested party during rule making. If the agency fails to respond to the request or refuses to accept the findings of the study because it could not get access to the original data, a court can determine whether there is cause for action. Such a determination can be made without having access to the underlying data or without creating an environment that encourages harassment, as is feared the Shelby Amendment does.
The perspective of the U.S. Chamber of Commerce (William L. Kovacs). For the scientific community to simply “fight Shelby,” said Mr. Kovacs, would be to “abdicate its responsibility. We need to take a step back, . . . and look at ways in which we can work together on those regulations that have an impact on public policy.”
The Chamber representative deplored the current number of federal regulations applicable to business. He said that there currently exist some 150,000 pages of federal regulations; that 4,000 new regulations are issued each year; and that regulation costs the economy $725 billion a year. This, he estimated, is the equivalent of three times the total value of the taxes paid by all U.S. companies.11 He urged greater sharing of any data that is instrumental in setting public policy, and emphasized that industry is not
interested in obtaining trade secrets, commercial information, or the identity of individuals.
Further, he commented that “maybe Shelby is too broad” but asserted that “no disclosure is not very good for democracy.” Rather than calling for “all data,” Mr. Kovacs suggested limiting the OMB requirement to data used in the development of public policy and he agreed with the notion of expanding the Shelby Amendment to apply to federal contracts as well.
The Chamber’s representative was asked about the HEI’s reassessment of the Six Cities Study. He listed several objections. The first was that the Chamber had filed a FOIA request with the EPA for “original data,” which it did not get. Second, the Chamber questioned the initial statement by the EPA that the new regulation would save 40,000 lives at $45 billion to $47 billion annually. “Whether it’s $45 billion a year or $150 billion a year,” said the representative, “we felt that when you have something which is going to put major areas of the country in non-attainment, that permits aren’t going to be permitted, businesses aren’t going to be expanded, it’s going to affect everything . . . , we thought that we should have a right to the data.” He noted that the denial of the Chamber’s FOIA requests was “the predicate for the lawsuit” that may be filed by the Chamber.
The perspective of a lobbying organization (Jim J. Tozzi-Center for Regulatory Effectiveness). According to Mr. Tozzi, the House of Representatives and Senate “leadership” set up the Center for Regulatory Effectiveness (CRE) in 1996 as part of the Congressional Review Act. The Center’s top priority is to ensure that high-quality data are used in setting regula
tory standards, noted Mr. Tozzi. Access to data is considered a prerequisite to assessing the quality of data used in regulatory decision making.
In recent years, the federal government has concluded that the efforts of various groups to oppose federal regulations in their traditional form had virtually halted the federal regulatory process. As a result, agencies began “off-register regulation”; that is, actions taken outside the Federal Register through appropriations bills, litigation, the release of information, and other means. For example, the Internet was used to publicize studies by federal agencies, which were then interpreted as official agency policy. Such practices were among the stimuli for the newly proposed “data quality legislation.”
Data quality legislation was passed in 2000 as an amendment to the Omnibus Appropriations Act. It specifies that the OMB must issue regulations to define minimum thresholds of data quality that can be disseminated by the federal government. It also declares that the agencies, after the OMB’s action, must issue their own data quality regulations.